ADLC

Theory & Introduction

How the ADLC phases and model failure modes map to the toolkit, with deep links to the original series.

Theory & Introduction

ADLC starts from one claim: agentic development should defend against model failure modes, not human ones. The full argument lives in the ADLC series — this page is the map.

The eight phases, two human gates

ADLC lifecycle — read the theory for P0

Eight phases (P0–P7), two human gates (P1 and P6), deterministic checks between everything. Read Two Human Gates and Everything Between Is Machine-Checked.

The model failure modes

F1Premature satisfaction (theory ↗)
F2Sycophancy (theory ↗)
F3Context rot (theory ↗)
F4Confident hallucination (theory ↗)
F5Reward hacking (theory ↗)
F6Finding-count prior (theory ↗)
F7Generative bloat (theory ↗)
F8Coherence loss (theory ↗)

Every phase, gate, and loop traces to one of these — see Stop Running the SDLC on Models That Aren't Human.

Where to go next

On this page